XEDAR As a Putative Colorectal Tumor Suppressor That Mediates P53-Regulated Anoikis Pathway

Oncogene (2009) 28, 3081–3092 & 2009 Macmillan Publishers Limited All rights reserved 0950-9232/09 $32.00 www.nature.com/onc ORIGINAL ARTICLE XEDAR as a putative colorectal tumor suppressor that mediates p53-regulated anoikis pathway

C Tanikawa1,3, Y Furukawa2, N Yoshida3, H Arakawa4, Y Nakamura1 and K Matsuda1

1Laboratory of Molecular Medicine, Human Genome Center, The University of Tokyo, Tokyo, Japan; 2Division of Clinical Genome Research, The University of Tokyo, Tokyo, Japan; 3Division of Gene Expression and Regulation, Institute of Medical Science, The University of Tokyo, Tokyo, Japan and 4Cancer Medicine and Biophysics Division, National Cancer Center Research Institute, Tokyo, Japan

Colorectal cancers with mutations in the p53 gene have an Introduction invasive property, but its underlying mechanism is not fully understood. Through the screening of two data sets Identification and characterization of cancer-related genes of the genome-wide expression profile, one for p53- are critical steps for the understanding of carcinogenic introduced cells and the other for the numbers of cancer mechanisms. Among the cancer-related genes that have tissues, we report here X-linked ectodermal dysplasia beenidentifiedsofar,inactivationofthep53 gene is the receptor (XEDAR), a member of the TNFR superfamily, most common alteration observed in human cancers as a novel p53 target that has a crucial role in colorectal (Beroud and Soussi, 2003; Hollstein et al., 1994). In carcinogenesis. p53 upregulated XEDAR expression response to various types of cellular stress, including DNA through two p53-binding sites within intron 1 of the damage, aberrant growth signal and oxidative stress, the XEDAR gene. We also found a significant correlation p53 protein is stabilized and accumulated. Activated p53 between decreased XEDAR expressions and p53 gene regulates many target genes that induce cell-cycle arrest, mutations in breast and lung cancer cell lines (P ¼ 0.0043 apoptosis, DNA repair and cellular senescence (Levine, and P ¼ 0.0122, respectively). Furthermore, promoter 1997; Vogelstein et al., 2000). We have isolated a number hypermethylation of the XEDAR gene was detected of p53 target genes, including p53AIP1, p53R2 and in 20 of 20 colorectal cancer cell lines (100%) and in 6 p53RDL1 (Nakamura, 2004; Oda et al., 2000; Tanaka of 12 colorectal cancer tissues (50%), respectively. Thus, et al., 2000; Tanikawa et al., 2003), and implicated the the XEDAR expression was suppressed to o25% of molecular mechanisms by which p53 regulated cell fate, surrounding normal tissues in 12 of 18 colorectal cancer death or survival, by balancing the expression levels of tissues (66.7%) due to either its epigenetic alterations these genes. However, an entire picture of the p53 signaling and/or p53 mutations. We also found that XEDAR pathway has not been disclosed yet. interacted with and subsequently caused the accumulation In this study, to identify a p53 target gene(s) that is of FAS protein, another member of p53-inducible TNFR. indispensable for p53-dependent tumor suppression, we Moreover, XEDAR negatively regulated FAK, a central used two data sets of the genome-wide expression profile component of focal adhesion. As a result, inactivation of obtained by cDNA microarray consisting of 36 864 XEDAR resulted in the enhancement of cell adhesion and cDNA fragments. One data set was obtained using the spreading, as well as resistance to p53-induced apoptosis. cells in which wild-type p53 was exogenously introduced, Taken together, our findings showed that XEDAR is a and the other was obtained using more than 1000 clinical putative tumor suppressor that could prevent malignant cancer cases (Kidokoro et al., 2008; Kitahara et al., 2001). transformation and tumor progression by regulating Through the analysis of these two data sets, we identified apoptosis and anoikis. XEDAR (X-linked ectodermal dysplasia receptor, also Oncogene (2009) 28, 3081–3092; doi:10.1038/onc.2009.154; known as EDA2R and TNFRSF27) as a novel p53 target, published online 22 June 2009 which mediated important p53 functions. The X-linked ectodermal dysplasia receptor is a Keywords: XEDAR; P53; FAS; colorectal member of the TNFR superfamily that is divided into cancer; anoikis; TNFR two subgroups because of difference in their cytoplasmic region. One class of TNFR, a death receptor, contains a cytoplasmic death domain. Several death receptors, such as FAS, and four TRAIL receptors, DR4, DR5, DcR1 and DcR2, were shown to be regulated by p53 (Liu Correspondence: Assistant Professor K Matsuda, Laboratory of et al., 2005; Wu et al., 1997) and their physiological roles Molecular Medicine, Human Genome Center, Institute of Medical in carcinogenesis have been well characterized (Lee Science, The University of Tokyo, 4-6-1 Shirokanedai, Minato, et al., 1999; Takakuwa et al., 2002). XEDAR belongs to Tokyo 108-8639, Japan. E-mail: [email protected] the other class of TNFR that lacks a discernible death Received 21 October 2008; revised 11 March 2009; accepted 12 May domain. This class of TNFRs interacts with TRAFs 2009; published online 22 June 2009 (TNFR-associated factors) and activates the nuclear XEDAR as a p53-induced regulator of anoikis C Tanikawa et al 3082 factor-kB (NF-kB) signaling, and consequently pro- mutant glioblastoma cells that were infected with motes cell proliferation. On the other hand, some adenovirus designed to express wild-type p53 (Ad-p53) members of this subclass were indicated to be involved or LacZ (Ad-LacZ) (Tanikawa et al., 2003). Thus, we in apoptotic pathways (Afford et al., 1999). Thus, the found a total of 60 novel p53 target genes that were TNFR superfamily is involved in various signaling upregulated by the exogenous introduction of wild-type pathways, including immune response, inflammation, p53. We then examined the expression profile database development and carcinogenesis. constructed by the same set of cDNA microarray using EDA-A1 and EDA-A2 are two major splicing iso- various cancers (Kitahara et al., 2001) and selected forms of EDA, and EDA-A2 specifically binds to XEDAR for further biological analysis because of its XEDAR (Yan et al., 2000). EDA-A1 binds to EDAR frequent downregulation in colorectal cancer tissues. and has an essential role for proper formation of skin To validate its regulation by p53, we carried out appendages, as the mutations of EDA, EDAR or its quantitative real-time PCR analysis and northern blot adaptor protein EDARADD were shown to cause analysis and found that the XEDAR expression was hypohidrotic ectodermal dysplasia (Smahi et al., 2002). remarkably induced by the introduction of p53 but not XEDAR is highly expressed in epidermal tissues during by that of LacZ (Figure 1a). Moreover, XEDAR protein embryogenesis (Yan et al., 2000), and mice that lacked was increased by Ad-p53 infection in a dose-dependent TRAF6, an adaptor protein of XEDAR, also displayed manner (Supplementary Figure 1). We also investigated hypohidrotic ectodermal dysplasia (Naito et al., 2002). the induction of XEDAR by DNA damage using MCF7 However, any mutations in the XEDAR gene have not (breast cancer) and A549 (lung cancer) cells with wild- been reported in individuals with hypohidrotic ectoder- type p53. We found that Adriamycin treatment remark- mal dysplasia, and XEDAR-deficient mice were indis- ably induced the XEDAR expression in both cells, tinguishable from their wild-type littermates at birth indicating the p53-dependent regulation of XEDAR (Newton et al., 2004). Thus, the physiological function expression (Figure 1b). of XEDAR has not been well clarified so far. Subsequently, we surveyed the genomic sequence of In this study, we report that XEDAR is frequently the XEDAR gene that is located on chromosome Xq12 inactivated in human colorectal cancers, and its and found two putative p53-binding regions (p53BR1 inactivation caused resistance to p53-induced apoptosis and p53BR2) within the first intron (Figure 1c). To and enhancement of cell adhesion. Thus, our findings examine the possible binding of p53 to these DNA suggested the crucial role of XEDAR in the anoikis segments, we carried out a chromatin immunoprecipita- pathway. In a multistep genetic model for colorectal tion (ChIP) assay using U373MG cells that were cancer, p53 mutations are more commonly found in infected with either Ad-p53 or Ad-LacZ. A PCR invasive colon cancer tissues (Vogelstein et al., 1989), analysis of immunoprecipitated DNA indicated that but the mechanisms by which p53 inhibits metastasis the p53 protein bound to the genomic fragment, have not been fully elucidated (Ilic et al., 1998; including p53BR1 (Figure 1c). We then subcloned a Nikiforov et al., 1996). We show the novel mechanism DNA fragment of 466 base pairs corresponding to that p53 suppresses colorectal carcinogenesis and tumor p53BR1, which included four putative p53-binding progression by regulating a novel p53 target, XEDAR. sequences (BS-A to D, respectively, Figure 1d) into the pGL3 promoter vector (pGL3/p53BR1) (Promega, Madison, WI, USA). We found that the co-transfection Results of pGL3/p53BR1 with wild-type p53 expression plasmid enhanced the luciferase activity more than 40-fold, Identification of XEDAR as a p53 target gene whereas the base substitutions within BS-A and BS-B To fully uncover p53 target genes, we examined a total segments completely diminished the luciferase activity of 36 864 cDNA fragments by means of cDNA (Figures 1c and d). The result of ChIP analysis suggested microarray using mRNAs isolated from U373MG p53 the weak association of p53 with p53BR2, but co-

Figure 1 Identiﬁcation of XEDAR as a novel p53 target gene. (a) Shows the quantitative PCR (qPCR) analysis (upper) and northern blot analysis (lower) of X-linked ectodermal dysplasia receptor (XEDAR) transcript in U373MG cells at indicated times after infection with Ad-p53 or Ad-LacZ at 8 multiplicity of infection (MOI). b2-Microglobulin and b-actin were used for the normalization of expression levels. p21WAF1 was served as a positive control. (b) Shows the qPCR analysis (upper) and western blot analysis (lower) of XEDAR at 48 h after treatment with adriamycin (ADR) in MCF7 and A549 cells. b2-Microblobulin and b-actin were used for the normalization of expression levels. (c) Represents the genomic structure of the XEDAR gene (upper). Black boxes indicate the locations and relative sizes of seven exons. The arrows indicate the potential p53-binding regions (p53BR1 and p53BR2). Chromatin immunoprecipitation (ChIP) assay was carried out using U373MG cells that were infected with Ad-p53 (lanes 1, 3–5) or Ad-LacZ (lane 2) (middle). DNA–protein complexes were immunoprecipitated with an anti-p53 antibody (lanes 2 and 3) followed by PCR ampliﬁcation. Input chromatin represents a portion of the sonicated chromatin before immunoprecipitation (lane 1). Immunopre- cipitates with an anti-Flag antibody (lane 4) or in the absence of antibody (lane 5) were used as negative controls. Results of luciferase assay of p53BR1 and p53BR2 are shown (lower). Luciferase activity is indicated relative to the activity of mock vector. (d) Shows the genomic structure of p53BR1 (upper). The arrows indicate the locations of p53BSs (BS-A to D) in p53BR1. Comparison of each p53BS with the consensus sequence (middle). R, purine; W, A or T; Y, pyrimidine. Identical nucleotides to the consensus sequence are written in capital letters. The underlined cytosine and guanine were substituted for thymine to introduce mutation at each p53-binding site. Results of luciferase assay of p53BR1 with or without mutations at either of p53BS are shown (lower). Luciferase activity is indicated relative to the activity of mock vector.

Oncogene XEDAR as a p53-induced regulator of anoikis C Tanikawa et al 3083 MCF7 A549 ) 3.5 -3 ) ) -3 3.0 35 -2 10 30 9 2.5 8 2.0 25 7 20 6 5

mRNA (x10 1.5 15 mRNA (x10 mRNA (x10 4 1.0 10 3 0.5 2 5 1

XEDAR 0 0 0 XEDAR 0 6 12 24 48 6 12 24 48 (h) 0 0.1 0.2 0.5 1 2 (μg/ml) XEDAR 0 0.1 0.2 0.5 1 2 (μg/ml) Ad-LacZ Ad-p53 ADR ADR Ad-p53 Ad-LacZ ADR ADR 0 6 12 24 48 0 6 12 24 48 (h)

0 0.2 1 (μg/ml) 00.21(μg/ml) XEDAR XEDAR XEDAR p21WAF1 p53 p53

β-actin β-actin β-actin

100 bp

p53BR1 p53BR2 5 kb BS-A BS-B BS-C BS-D

p53BR1 466 bp exon1 exon2 exon7 mt BS-A mt BS-B

Ad-LacZ Ad-p53 mt BS-C mt BS-D mt BS-A,B

– IP p53BS RRRCWWGYYYRRRCWWGYYY Input chromatin anti-p53 anti-p53 anti-FLAG

p53BR1 BS-A GAACATGCCTGGACgTGTCC BS-B G GGCAAGTC aGGGCT TGT g T p53BR2 BS-C GAGCTTGgTT t cACTTGaTC BS-D GG c CAAGaCCcAGCAAGCTg p21WAF1

60 Mock Mock 60 50 wt-p53 p53 50 mt-p53 40 40

30 30 20 20 10 10 0 Relative luciferase activity (folds) Relative luciferase activity (folds) 0 p53BR1 p53BR2 pGL3 p53BR1 mt BS-B mt BS-A mt BS-C mt BS-D mt BS-A,B

Oncogene XEDAR as a p53-induced regulator of anoikis C Tanikawa et al 3084 transfection of p53 with pGL3/p53BR2 did not enhance contribute to the reduced XEDAR expression in color- the luciferase activity (Figure 1c). These findings ectal cancer cells. We also found the partial restoration indicated that p53 directly regulated the XEDAR of XEDAR protein expression in 5-aza-20-deoxycyti- expression through two p53-responsible elements, dine-treated HCT116 cells (Supplementary Figure 2). BS-A and BS-B. We further measured quantitatively the XEDAR expression in colorectal cancer tissues and its correspond- Expression of XEDAR was frequently suppressed in ing normal tissues microdissected from 18 frozen clinical colorectal cancer samples. We found its decreased expression (o25% of We then analyzed the expression level of XEDAR in 83 its corresponding normal tissue) in 12 colorectal cancer cell lines (20 of colorectal cancer, 22 of breast cancer, 35 tissues (66.7%) (Figure 2e). of lung cancer and 6 control cell lines) and 26 normal We then examined genomic DNAs from 68 colorectal adult tissues (Supplementary Table 2). We also exam- cancer tissues and 20 colorectal cancer cell lines. We ined cDNA sequences of p53 using mRNAs isolated identified one somatic mutation at the splice acceptor from these cancer cells. As we expected, XEDAR site (IVS3 À1 G>A) in one female clinical colorectal expressions were significantly reduced in p53 mutant cancer case and one missense mutation (exon 2 A74C; breast and lung cancer cells compared with p53 wild- Y8H) in one colorectal cancer cell line, NCI-H716 type cancer cells (P ¼ 0.0043 and P ¼ 0.0122, respec- (derived from a male patient) (Figure 2f). These base tively), as well as 6 normal cell lines and 26 normal adult substitutions were found in none of the normal control tissues (Figure 2a). In addition, when we treated tissues in the 68 patients or 96 normal healthy HEK293 and NHDF cells with siRNA oligonucleotide individuals examined. The splicing site mutation would designed to suppress p53 (sip53), we found that cause the aberrant splicing of XEDAR transcript and XEDAR expression levels were remarkably decreased dysfunction of the XEDAR protein. The data shown (Figure 2b). These data clearly implicated that the above indicated a possible role of XEDAR as a inactivation of p53 would cause the downregulation of colorectal tumor suppressor. XEDAR. However, for colorectal cancer cell lines, XEDAR expressions were reduced even in the cells Role of XEDAR in anoikis without p53 mutations. Therefore, we considered there We then carried out immunocytochemical analysis and might be another mechanism that suppressed the found that the ectopically expressed XEDAR was XEDAR expression. Transcriptional silencing of tumor accumulated at the leading edge in cells that exhibited suppressor genes by DNA hypermethylation is a process formation (Figure 3a). As the p53 gene mutation common epigenetic event in malignancies. We was associated with invasive colorectal cancer, we sequenced genomic DNA isolated from colorectal suspected that the inactivation of XEDAR might cause cancer cell lines and clinical tissues after bisulfate anchorage independency. To further investigate the role treatment and found tumor-specific hypermethylation of XEDAR in cell adhesion or cell attachment, we of CpG islands (À190/ þ 73) in all of the 20 cancer cell transfected HEK293 cells with siRNA oligonucleotides lines and in half of the colorectal cancer tissues obtained designed to suppress XEDAR (siXEDAR) or EGFP from 12 male patients (Figure 2c). We then treated seven (siEGFP) for 48 h and resuspended and plated in fresh colorectal cancer cell lines with the demethylating agent culture dishes, as previously described (Figure 3b) 5-aza-20-deoxycytidine and observed a significant re- (Liang et al., 2007). We investigated the morphological storation of the XEDAR expression in three of the four change of each cell and found that siXEDAR-treated colorectal cancer cell lines with wild-type p53 back- HEK293 cells indicated increased cell adhesion (4 h after ground but not in any of the three p53 mutant cell lines plating) and process formation (12 h) compared with the (Figure 2d), indicating that epigenetic alternations control cells (Figure 3b). These findings suggested the

Figure 2 XEDAR as a colorectal tumor suppressor. (a) Boxplots of X-linked ectodermal dysplasia receptor (XEDAR) expression in 83 cell lines; 20 of colorectal cancer (CRC), 22 of breast cancer (BC), 35 of lung cancer (LC), 6 of normal cell lines (N), and 26 normal tissues (NT). Student’s t-test was applied for comparing the XEDAR expressions in p53 mutant cell lines with those in p53 wild-type cell lines. XEDAR expression was determined by quantitative PCR (qPCR) analysis. b2-Microglobulin was used for the normalization of expression levels. (b) qPCR analysis (upper) and western blot analysis (lower) of XEDAR at 48 h after the transfection with siRNA oligonucleotide designed to suppress p53 expression. EGFP was used as control. b2-Microglobulin and b-actin were used for the normalization of expression levels. (c) Schematic representation of 50-flanking region of the XEDAR gene (upper). Black box indicates first exon. Vertical bars indicate CpG sites. Regions analyzed by direct bisulfite sequencing are shown by black bar below the CpG sites (À190/ þ 73). DNA sequencing analysis after bisulfite modification was carried out in 20 colorectal cancer cell lines (lower, left) and 12 pairs of colorectal cancer tissues from male patients (lower, right). The methylation status of 13 CpG sites was examined. Closed boxes indicate methylated sites, gray and open boxes indicate partially methylated and unmethylated sites, respectively. The p53 mutation status of colorectal cancer cell lines was also indicated; wild type (wt) and mutant (mt). (d) Semi-quantitative real time (RT)–PCR analysis after treatment with the demethylating agent 5-aza-20-deoxycytidine in several colorectal cancer cell lines. b2-Microglobulin was used for the normalization of expression levels. (e) Relative XEDAR expressions in colorectal cancer tissues compared with its surrounding normal tissues were examined by qPCR methods. b2-Microglobulin was used for the normalization of expression levels. (f) Mutations of the XEDAR gene in colorectal cancer tissue and colorectal cancer cell line. Colorectal cancer tissue (CRC66T) has a G to A substitution at the splice acceptor site. NCI-H716 has a T to C substitution in exon 2 (Tyr8His). Sequences of the corresponding normal tissue (CRC66N) and the control DNA are also shown. Genomic structure and domain structure of XEDAR are shown at the lower panel. The asterisks indicate the location of mutations.

Oncogene XEDAR as a p53-induced regulator of anoikis C Tanikawa et al 3085 inhibitory effects of XEDAR on cell adhesion and/or cancer cells carrying mutation in the XEDAR gene were motility. The presence of the p53 mutation was non-adherent cells that acquired anchorage indepen- indicated to be signiﬁcantly correlated with metastasis dency, whereas the remaining 19 colorectal cancer cells and poor prognosis of various cancers (Diez et al., 2000; with wild-type XEDAR were adherent cells. Hence, we Pharoah et al., 1999). Interestingly, NCI-H716 colon suspected that loss of functional XEDAR might have a

P = 0.2121 P = 0.0043 P = 0.0122

1.0E-02 5-aza-2’-deoxycytidine 0 0.01 0.05 0.2 1 0 0.01 0.05 0.2 1 (µM)

1.0E-04 HCT116 mRNA

LoVo 1.0E-06 p53 XEDAR wt LS174T

1.0E-08 RKO n 13 7 16 6 27 8 6 26 p53 status mt wt mt wt mt wt DLD1

CRC BC LC N NT p53 HCT15 mt HEK293 NHDF KM12C 16 2.5 ) ) -2 -2 14 2.0 XEDAR B2M 12 10 1.5 1.0E+03 8 mRNA (x10 mRNA (x10 6 1.0 1.0E+02 4 0.5 2 1.0E+01 XEDAR XEDAR

0 0 (T/N) 1.0E+00 EGFP p53 siRNA EGFP p53 siRNA 1.0E-01 XEDAR XEDAR XEDAR of 1.0E-02 Relative expression 1.0E-03 p53 p53 1.0E-04 123456789101112131415161718

-actin -actin CRC66 N CRC66 T control NCI-H716 exon4 100 bp **exon4 exon1 **

CpG

12345678910111213CpG 12345678910111213CpG DLD-1 mt N G A/G TAC = Y CAC = H HCT-15 mt T N HCT 116 wt CRC66 N IVS3 -1 G>A HT-29 mt T KM12C mt N 5 kb T KM12SM mt LoVo wt N T LS 174T wt N wt NCI-H498 T NCI-H508 mt N NCI-H716 NCI-H716 mt T Y8H RKO wt N 100 aa TNFR domain SNU-C2A mt T SNU C4 wt N Colorectal cancer cell lines SNU C5 mt T

Colorectal cancer tissues Transmembrane SW48 wt N domain SW480 mt T SW620 mt N SW948 mt T WiDr mt N T N T

Oncogene XEDAR as a p53-induced regulator of anoikis C Tanikawa et al 3086 key role in the metastatic property of p53-mutated endogenous FAS protein was significantly increased cancer. We then carried out a colony formation assay in (Figure 4b). Furthermore, knockdown of FAS also soft agar using cancer cell lines with low or absent suppressed the expression of XEDAR (Figure 4b). Co- expression of XEDAR (DLD-1, HCT116, SW480, immunoprecipitation or immunocytochemical analysis SW620 and H1299 cells) and found that XEDAR could showed a physiological interaction of XEDAR and FAS suppress anchorage-independent tumor cell growth at the plasma membrane (Figure 4c). These results (Figure 3c). Thereafter, we examined the involvement suggested that the XEDAR–FAS interaction was likely of XEDAR in detachment-induced apoptosis. Forty- to cause stabilization of both proteins and subsequently eight hours after transfection with plasmid expressing promote apoptotic pathway. XEDAR, HEK293 cells were detached and cultured in Finally, we examined the role of XEDAR in the p53- suspension for 24 h using a poly-2-hydroxyethyl metha- induced apoptosis. Introduction of p53 into U373MG crylate-coated plate (Folkman and Moscona, 1978). cells induced cell rounding and process retraction before Subsequent analyses indicated that XEDAR introduc- apoptotic cell death, whereas siXEDAR-treated cells tion reduced the number of viable cells through caspase- maintained spindle-shape morphology and subsequently 3 activation (Figure 3d). Thus, we showed the significant attenuated Ad-p53-induced apoptosis (Figure 4d). In role of XEDAR in detachment-induced apoptosis, addition, treatment with siXEDAR significantly inhib- namely anoikis. ited adriamycin-induced growth suppression in MCF7 cells (Figure 4d). Regulation of FAS and FAK by XEDAR in a Taken together, XEDAR controlled apoptosis signal- p53-dependent apoptotic pathway ing and cell adhesion through the functional interaction We then investigated the role of XEDAR in the p53- and regulation of FAS and FAK, and XEDAR downstream pathway. A recent analysis indicated that inactivation possibly resulted in the invasive property p53 suppressed FAK expression (Golubovskaya et al., of p53-mutant cancer cells. Our findings showed a novel 2008), one of the essential components in focal adhe- mechanism that prevents malignant transformation sion. Ectopic expression of FAK could enhance cell and tumor progression by the p53/XEDAR pathway adhesion and attachment in various cancer cells (Ilic (Figure 5). et al., 1995) similar to the XEDAR knockdown cells. Therefore, we examined the effect of XEDAR on FAK expression. Interestingly, introduction of XEDAR Discussion remarkably suppressed FAK expression (Figure 3d), and downregulation of XEDAR in Ad-p53-infected Our microarray analysis identified dozens of unchar- U373MG cells or H1299 cells partially diminished FAK acterized possible p53 target genes, which might mediate suppression by p53 (Figure 4a). These results indicated important p53 functions. Some p53 target genes, such as the role of XEDAR in the negative regulation of FAK p21 and BAX, which have significant roles in p53- by p53. dependent tumor suppression were downregulated in As XEDAR was shown to induce apoptosis through cancer tissues (Furutani et al., 1997; Rampino et al., the activation of caspase-8, which is a key mediator of 1997). Therefore, the expression analysis of candidate FAS-induced apoptotic signaling (Sinha and Chaudh- p53 target genes in cancer tissues offers valuable ary, 2004), we investigated the physiological and information for their possible roles in human carcino- functional interactions between FAS and XEDAR. genesis. Downregulation of XEDAR in p53-infected U373MG XEDAR is highly expressed in embryonic tissues, and cells remarkably suppressed FAS induction (Figure 4a), the mutation of the EDA gene that encodes its ligand whereas p21 expression was not affected. A similar EDA-A2 caused hypohidrotic ectodermal dysplasia. In result was observed in XEDAR-suppressed HEK293 addition, we found that XEDAR was a direct transcrip- and NHDF cells (Figure 4b and Supplementary tional target of p63, a member of the p53 family, which Figure 3). In addition, when XEDAR protein was has an important role in epidermal development (data overexpressed in HEK293 cells, the expression of not shown). However, the evidence obtained so far

Figure 3 XEDAR as a mediator of detachment-induced apoptosis. (a) HEK293 cells were transfected with X-linked ectodermal dysplasia receptor (XEDAR) expression plasmid. At 24 h after transfection, the cells were detached and seeded on a fresh plate. After 60 h of incubation, the cells were fixed and double stained with anti-XEDAR antibody (Alexa fluor 488) and Alexa fluor 594 phalloidin to visualize actin filaments. (b) At 48 h after transfection of each siRNA into HEK293 cells, the cells were detached and seeded on a fresh plate. The expression of XEDAR was shown (left). After 4 h (spreading assay, middle) or 12 h (processing morphogenesis, right) of the incubation, the proportion of spreading or processing cells was indicated (upper panels). The representative images of spreading or processing cells were shown (lower panels, arrowheads). (c) Cells were transfected with either of two plasmids expressing the sense strand (SE, pcDNA3.1 þ /XEDAR) or the antisense strand (AS, pcDNA3.1-/XEDAR) of XEDAR, and colony formation assay was carried out in soft agar. The expression of XEDAR was shown (left). The cells were cultured in the presence of geneticin (0.4, 0.5, 1.0, 1.0 and 0.8 mg/ml for DLD-1, HCT116, SW480, SW620 and H1299 cells, respectively) for 2 weeks. Numbers of colonies generated were quantified using the Image J software. (d) Twenty-four hours after the transfection with pcDNA3.1 þ /XEDAR or mock plasmid, HEK293 cells were seeded in a poly-2-hydroxyethyl methacrylate-coated plate. The viability, caspase-3 activity and FAK expression in the cells were analyzed after the 24-h incubation. Results are given as ratio against the cells transfected with mock plasmid. b-Actin was used for the normalization of the expression levels. *Po0.05 by Student’s t-test.

Oncogene XEDAR as a p53-induced regulator of anoikis C Tanikawa et al 3087 failed to clarify the role of XEDAR in development. The X-linked ectodermal dysplasia receptor (XEDAR) In this study, we showed the crucial role of the p53/ interacts with TRAF3 and TRAF6, activates NF-kB XEDAR pathway in human carcinogenesis. signaling and consequently promotes cell proliferation

XEDAR Phalloidin Merge

Spreading assay Processing morphogenesis

35 * 70 * 30 60 25 50 20 40 siRNA kDa EGFP XEDAR 15 30 10 20 XEDAR 50

Spreading cells (%) 5 10 Processing cells (%) 0 0 -actin EGFP XEDAR siRNA EGFP XEDAR siRNA

DLD-1 HCT116 SW480 SW620 H1299 H1299 1.2 1.2 1.2 1.2 1.2 SE AS kDa 1.0 1.0 1.0 1.0 1.0

XEDAR 50 0.8 0.8 0.8 0.8 0.8 * 0.6 0.6 * 0.6 0.6 * 0.6 -actin 0.4 0.4 0.4 0.4 0.4 * 0.2 0.2 0.2 0.2 0.2 Relative number of colonies Relative number of colonies Relative number of colonies Relative number of colonies 0 Relative number of colonies 0 0 0 0 AS SE AS SE AS SE AS SE AS SE

120 1.8 * 1.6

100 Mock XEDAR 1.4 * 80 1.2 FAK 1.0 60 0.8 40 0.6 XEDAR

Viable cells (%) 0.4 20 0.2 -actin

0 Relative Activity of Caspase 3 0 Mock XEDAR Mock XEDAR

Oncogene XEDAR as a p53-induced regulator of anoikis C Tanikawa et al 3088 (Sinha et al., 2002). On the other hand, the EDA-A2/ either the p53 mutation or promoter hypermethylation XEDAR pathway was also shown to promote apoptotic is sufficient to suppress the XEDAR expression. In cell death through the DISC (death-inducing signaling addition, we found, although rare, genetic mutations complex) formation containing FADD, caspase-8, that would cause the inactivation of XEDAR in caspase-10 and c-FLIP (Sinha and Chaudhary, 2004). colorectal cancers. Taken together, our findings indi- However, XEDAR lacks a discernible death domain; cated that XEDAR is a putative colorectal tumor the molecular mechanism of XEDAR-induced apopto- suppressor that would prevent the malignant transfor- sis was not well understood. Some TNFR members that mation and tumor progression by the cross-talk between lacked the death domain were shown to induce FAS and FAK. apoptosis through the interaction with FAS, another p53-inducible TNFR superfamily member (Afford et al., 1999; Grell et al., 1999). Similarly, our result suggested Materials and methods that XEDAR promoted apoptotic signaling through the interaction with FAS. We also examined the effect of cDNA microarray FAS and XEDAR on NF-kB signaling; however, cDNA microarray analysis was carried out as previously described (Kitahara et al., 2001; Mori et al., 2002). Briefly, coexpression of XEDAR and FAS did not enhance colorectal cancer tissues or its surrounding normal mucosa the NF-kB pathway (data not shown), indicating that that were obtained with informed consent were microdissected the XEDAR–FAS interaction would rather promote from frozen tissue sections. The total RNA of each sample was apoptotic signaling than NF-kB-mediated cell prolifera- isolated and amplified using RNeasy spin column kits (Qiagen, tion. Valencia, CA, USA) and T7-Transcription kit (Epicentre We also found that XEDAR negatively regulated Technologies, Madison, MI, USA). Replication-deficient FAK, which is a major mediator of cell adhesion and recombinant adenovirus encoding p53 (Ad-p53) or LacZ functions as an adaptor protein that transduces adhe- (Ad-LacZ) was generated and purified, as previously described sion-dependent and growth factor-dependent signaling (Oda et al., 2000). U373MG cells were infected with viral (McLean et al., 2005). FAK expression was frequently solutions at an indicated multiplicity of infection and incubated at 37 1C until the time of harvest. poly(A) þ RNAs increased in various cancer tissues, including colorectal were isolated from U373MG cells using a standard protocol. cancer, and its upregulation was associated with poor Each RNA sample was labeled and hybridized to a microarray prognosis and anoikis resistance (Lark et al., 2003; consisting of 36 864 cDNA fragments. The microarray Owens et al., 1995; Weiner et al., 1993). Metastasis data set is accessible from the Gene Expression Omnibus occurs through a series of sequential steps in which (http://www.ncbi.nlm.nih.gov/geo/index.cgi, Accession No. tumor cells first migrate from the primary tumor, GSE14953). penetrate blood vessels and colonize distant sites. Among them, the metastatic potential of tumors is Cell culture and transfections Each cell line was purchased from the American Type Culture generally associated with an increased resistance to Collection (Manassas, VA, USA), Lonza Biologics (Ports- anoikis during the initial step of cell migration. Our mouth, NH, USA) or JCRB (Osaka, Japan). Cells were findings indicated that XEDAR inactivation would transfected with plasmids using FuGENE6 (Roche, Basel, cause FAS suppression and FAK activation and Switzerland) or Lipofectamine LTX (Invitrogen, Carlsbad, subsequently confer anoikis resistance in cancer cells. CA, USA). siRNA oligonucleotides, commercially synthesized We also found three mechanisms that inactivated by Sigma Genosis (St Louis, MO, USA), were transfected with XEDAR in colorectal cancer cells. One mechanism was Lipofectamine 2000 reagent (Invitrogen) for 4 h. Sequences for p53 mutations that were found in nearly half of the each oligonucleotide are indicated in Supplementary Table 1. colorectal cancers (Beroud and Soussi, 2003). The For suspension culture, 6-well plates were coated with 2 ml of second one was inactivation by the promoter hyper- poly-HEMA (poly-2-hydroxyethyl methacrylate; Sigma, St Louis, MO, USA) solution in ethanol (10 mg/ml) for at least 3 methylation that was also observed in half of the days until the solvent had evaporated completely. The cells colorectal cancer tissues and in all of the colorectal were suspended in 2 ml of media containing 0.5% methylcel- 0 cancer cell lines. The fact that treatment with 5-aza-2 - lulose and plated onto poly-HEMA-coated dishes. For deoxycytidine restored XEDAR expression in none of methylation analysis, the cells were treated with an indicated the p53 mutant colorectal cancer cell lines indicated that dosage of 5-aza-20-deoxycytidine (Sigma).

Figure 4 X-linked ectodermal dysplasia receptor (XEDAR) as a mediator of p53-dependent apoptosis. (a) Expressions of endogenous XEDAR, FAS, FAK and p21WAF1 were examined in U373MG or H1299 cells 48 h after infection with Ad-p53. Each siRNA was transfected 6 h before infection with Ad-p53. (b) Expression of XEDAR and FAS protein in HEK293 cells after transfection with each siRNA (left) or plasmid (right). b-Actin was used for the normalization of expression levels. (c) Cell lysates from HEK293T cells that were transfected with plasmids expressing either Flag-XEDAR and/or HA-FAS were immunoprecipitated using anti-Flag or anti-HA antibody, followed by immunoblotting with anti-Flag and anti-HA antibody, respectively (left). Subcellular localization of XEDAR and FAS was examined by immunocytochemistry (right). Cells were transfected with two plasmids expressing either Myc-XEDAR or HA-FAS. Cells were double stained with anti-Myc antibody (Alexa ﬂuor 488) and anti-HA antibody (Alexa ﬂuor 594). (d) At 6 h after transfection of each siRNA, U373MG cells were infected with Ad-p53. Representative images of Ad-p53 infected cells were shown (left). siEGFP was used as control. Analysis of p53-induced apoptotic cell death in XEDAR knockdown cells (middle) is shown. siXEDAR or siEGFP was transfected into U373MG cells 6 h before infection with Ad-p53 or Ad-LacZ. Proportions of apoptotic cells are indicated as a percentage of sub-G1 fractions in FACS analysis. Each siRNA was transfected into MCF7 cells 6 h before treatment with adriamycin (ADR) (right). Cell viability was examined by MTT assay. *Po0.05 by Student’s t-test.

Oncogene XEDAR as a p53-induced regulator of anoikis C Tanikawa et al 3089 Quantitative real-time PCR ChIP assay Quantitative real-time PCR was conducted using the SYBR Chromatin immunoprecipitation assay was carried out using Green I Master or Probe Master on a LightCycler 480 the CHIP Assay kit (Upstate Biotechnology, Waltham, MA, (Roche). The primer and probe sequences are indicated in USA) as previously described (Tanikawa et al., 2003). PCR Supplementary Table 1. The mRNA of 26 normal tissues was ampliﬁcations of XEDAR intron 1, containing the consensus purchased from TAKARA Clontech (Kyoto, Japan). p53-binding sites, were performed on immunoprecipitated

Ad-p53 - + - +

Mock XEDAR siRNA EGFP XEDAR EGFP XEDAR EGFP XEDAR EGFP XEDAR siRNA XEDAR EGFP FAS 60 36 0 36 60 (h)

XEDAR FAS FAS

FAS XEDAR XEDAR

FAK -actin -actin

p21WAF1

-actin

U373MG H1299

DAPI XEDAR Flag-XEDAR - + - + - + - + HA-FAS - - + + - - + +

Input

IP : Anti-Flag

IP : Anti-HA

IB : Anti-Flag IB : Anti-HA

FAS Merge

70 * 0.8 ** 60 0.7 50 0.6 0.5 40 0.4 30 0.3

siEGFP 20 Absorbance 0.2

Apoptotic Cells (%) 10 0.1 0 0 siRNA siRNA EGFP EGFP EGFP EGFP EGFP XEDAR XEDAR XEDAR XEDAR XEDAR ( Ad-LacZ Ad-p53 0 1 2 g/ml)

siXEDAR

Oncogene XEDAR as a p53-induced regulator of anoikis C Tanikawa et al 3090 Methylation ethanol precipitated and resuspended in double-distilled water. Mutation p53 Mutation The modified DNA was subjected to PCR amplification of the CpG islands in the XEDAR promoter using the primers indicated in Supplementary Table 1. Amplified products were subcloned using the TOPO-TA Cloning System (Invitrogen). Plasmid DNA of at least six insert-positive clones was isolated XEDAR and sequenced using the ABI sequencing system (Applied Biosystems, Foster City, CA, USA).

Mutation analysis The entire coding region of the p53 gene was amplified by PCR using cDNA prepared from 77 cancer cell lines, and PCR XEDAR products were directly sequenced. For analysis in the XEDAR gene, genomic DNA was purified by standard protocol. Six coding exons of the XEDAR gene were amplified, purified and sequenced. The primers used in this analysis are indicated in Supplementary Table 1.

FAS FAK Antibodies Anti-Flag monoclonal (clone M2) and polyclonal (F7425) antibody, as well as anti-b-actin monoclonal antibody (clone AC15) were purchased from Sigma. Anti-p53 monoclonal WAF1 Apoptosis Cell adhesion antibody (Ab-12, clone DO-7) and anti-p21 monoclonal antibody (Ab-1, clone EA10) were purchased from Calbio- chem (San Diego, CA, USA). Anti-XEDAR polyclonal antibody (T-14), anti-FAS monoclonal antibody (B-10), anti- HA monoclonal (F-7) and polyclonal (Y-11) antibody, as well as anti-c-Myc polyclonal antibody (A-14) were purchased Anoikis from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Anti- Figure 5 A schema of X-linked ectodermal dysplasia receptor HA monoclonal antibody (clone 3F10) was purchased from (XEDAR) function as a tumor suppressor gene. XEDAR was Roche. Anti-FAK polyclonal antibody (ab2999) was pur- induced through a p53-dependent manner and mediated anoikis chased from Abcam (Cambridge, UK). Rabbits were immu- pathway through the cross-talk between FAS and FAK. However, nized with the recombinant proteins corresponding to the XEDAR is frequently inactivated in colorectal cancers by the p53 extracellular domain (amino acids 1–139) of XEDAR. Anti- mutation, or by its genetic or epigenetic alterations. bodies were subsequently purified on antigen affinity columns. For labeling F-actin, Alexa fluor 594 phalloidin (Molecular Probes, Eugene, OR, USA) was used.

chromatin using the primers indicated in Supplementary Spreading assay and processing morphogenesis Table 1. Forty-eight hours after transfection with each siRNA oligonucleotide, HEK293 cells were detached with 0.02% ethyle- Gene reporter assay nediaminetetraacetic acid (EDTA), resuspended in Dulbecco’s DNA fragments, including potential p53-binding sites of the modified Eagle’s medium (DMEM) with 10% of fetal bovine XEDAR gene, were amplified and subcloned into the pGL3- serum and plated onto 6-well plates. Spreading cells were promoter vector (Promega). The primers for amplification are counted after 4 h of incubation, and processing morphogenesis indicated in Supplementary Table 1. To make a series of was determined 12 h later. mutant vectors, a point mutation ‘T’ was inserted into the site of the fourth and the fourteenth nucleotide ‘C’ and into the Anchorage-independent growth assay seventh and the seventeenth nucleotide ‘G’ of the consensus Soft agar assays were carried out in 6-well culture plates. p53-BS using the QuickChange site-directed mutagenesis kit A volume of 2 ml of culture media with 0.5% agar was (Stratagene, La Jolla, CA, USA). A reporter assay was carried solidified in the bottom of each well. At 24 h after transfection out using the Dual Luciferase assay system (Promega) as with either plasmid, equal numbers of cells were suspended in previously described (Oda et al., 2000). 1.5 ml of media with 0.33% agar and added to each well. Cells were fed with 1 ml of media supplement with geneticin Bisulfite sequencing analysis (Invitrogen) every 3 days. After 2 weeks of incubation, Genomic DNA of 3 mg was digested for 16 h with 30 U of colonies were stained with iodonitrotetrazolium chloride Sau3AI (Takara, Tokyo, Japan) in a 150 ml of reaction volume. (Sigma) and scored using the Image J software. The digested DNA was denatured in 0.3 M of NaOH for 20 min at 37 1C, and then the unmethylated cytosine residues were Cell death assay sulfonated by incubation in 3.12 M of sodium bisulfite (pH 5.0) Cells were incubated with adriamycin for 2 h or infected with and 0.5 mM of hydroquinone at 55 1C for 16 h. The sulfonated either Ad-p53 or Ad-LacZ at 6 h after transfection of siRNA DNA was recovered using the QIAquick PCR purification oligonucleotide. Apoptotic cells were quantified by FACS system (Qiagen) according to the manufacturer’s recommen- analysis as previously described (Matsuda et al., 2002). dations. The conversion reaction was completed by desulfo- Activities of caspase-3 were monitored using a caspase-3 assay nating in 0.3 M of NaOH for 20 min at 37 1C. The DNA was kit (MBL, Nagoya, Japan). Cell viability was determined using

Oncogene XEDAR as a p53-induced regulator of anoikis C Tanikawa et al 3091 the MTT assay using Cell Counting Kit-8 (Dojindo, Kuma- Acknowledgements moto, Japan). We thank T Katagiri for helpful discussion and A Takahashi K Makino for her technical assistance. This work was supported partly by grant #18687012 from Japan Society for Conﬂict of interest the Promotion of Science and Ministry of education, culture, sports, science and technology of Japan (to KM). CT is a JSPS The authors declare no conﬂict of interest. Research Fellow.

References

Afford SC, Randhawa S, Eliopoulos AG, Hubscher SG, Young LS, Liu X, Yue P, Khuri FR, Sun SY. (2005). Decoy receptor 2 (DcR2) is a Adams DH. (1999). CD40 activation induces apoptosis in cultured p53 target gene and regulates chemosensitivity. Cancer Res 65: human hepatocytes via induction of cell surface fas ligand 9169–9175. expression and amplifies fas-mediated hepatocyte death during Matsuda K, Yoshida K, Taya Y, Nakamura K, Nakamura Y, allograft rejection. J Exp Med 189: 441–446. Arakawa H. (2002). p53AIP1 regulates the mitochondrial apoptotic Beroud C, Soussi T. (2003). The UMD-p53 database: new mutations pathway. Cancer Res 62: 2883–2889. and analysis tools. Hum Mutat 21: 176–181. McLean GW, Carragher NO, Avizienyte E, Evans J, Brunton VG, Diez M, Medrano M, Muguerza JM, Ramos P, Hernandez P, Villeta Frame MC. (2005). The role of focal-adhesion kinase in cancer— R et al. (2000). Influence of tumor localization on the prognostic a new therapeutic opportunity. Nat Rev Cancer 5: 505–515. value of P53 protein in colorectal adenocarcinomas. Anticancer Res Mori T, Anazawa Y, Iiizumi M, Fukuda S, Nakamura Y, 20: 3907–3912. Arakawa H. (2002). Identification of the interferon regulatory Folkman J, Moscona A. (1978). Role of cell shape in growth control. factor 5 gene (IRF-5) as a direct target for p53. Oncogene 21: Nature 273: 345–349. 2914–2918. Furutani M, Arii S, Tanaka H, Mise M, Niwano M, Harada T et al. Naito A, Yoshida H, Nishioka E, Satoh M, Azuma S, Yamamoto T (1997). Decreased expression and rare somatic mutation of the et al. (2002). TRAF6-deficient mice display hypohidrotic ectodermal CIP1/WAF1 gene in human hepatocellular carcinoma. Cancer Lett dysplasia. Proc Natl Acad Sci USA 99: 8766–8771. 111: 191–197. Nakamura Y. (2004). Isolation of p53-target genes and their functional Golubovskaya VM, Finch R, Kweh F, Massoll NA, Campbell- analysis. Cancer Sci 95: 7–11. Thompson M, Wallace MR et al. (2008). p53 regulates FAK Newton K, French DM, Yan M, Frantz GD, Dixit VM. (2004). expression in human tumor cells. Mol Carcinog 47: 373–382. Myodegeneration in EDA-A2 transgenic mice is prevented by Grell M, Zimmermann G, Gottfried E, Chen CM, Grunwald U, XEDAR deficiency. Mol Cell Biol 24: 1608–1613. Huang DC et al. (1999). Induction of cell death by tumour necrosis Nikiforov MA, Hagen K, Ossovskaya VS, Connor TM, Lowe SW, factor (TNF) receptor 2, CD40 and CD30: a role for TNF-R1 Deichman GI et al. (1996). p53 modulation of anchorage activation by endogenous membrane-anchored TNF. EMBO J 18: independent growth and experimental metastasis. Oncogene 13: 3034–3043. 1709–1719. Hollstein M, Rice K, Greenblatt MS, Soussi T, Fuchs R, Sorlie T et al. Oda K, Arakawa H, Tanaka T, Matsuda K, Tanikawa C, Mori T et al. (1994). Database of p53 gene somatic mutations in human tumors (2000). p53AIP1, a potential mediator of p53-dependent and cell lines. Nucleic Acids Res 22: 3551–3555. apoptosis, and its regulation by Ser-46-phosphorylated p53. Cell Ilic D, Almeida EA, Schlaepfer DD, Dazin P, Aizawa S, Damsky CH. 102: 849–862. (1998). Extracellular matrix survival signals transduced by focal Owens LV, Xu L, Craven RJ, Dent GA, Weiner TM, Kornberg L et al. adhesion kinase suppress p53-mediated apoptosis. J Cell Biol 143: (1995). Overexpression of the focal adhesion kinase (p125FAK) in 547–560. invasive human tumors. Cancer Res 55: 2752–2755. Ilic D, Furuta Y, Kanazawa S, Takeda N, Sobue K, Nakatsuji N et al. Pharoah PD, Day NE, Caldas C. (1999). Somatic mutations in the p53 (1995). Reduced cell motility and enhanced focal adhesion gene and prognosis in breast cancer: a meta-analysis. Br J Cancer contact formation in cells from FAK-deficient mice. Nature 377: 80: 1968–1973. 539–544. Rampino N, Yamamoto H, Ionov Y, Li Y, Sawai H, Reed JC et al. Kidokoro T, Tanikawa C, Furukawa Y, Katagiri T, Nakamura Y, (1997). Somatic frameshift mutations in the BAX gene in Matsuda K. (2008). CDC20, a potential cancer therapeutic target, is colon cancers of the microsatellite mutator phenotype. Science negatively regulated by p53. Oncogene 27: 1562–1571. 275: 967–969. Kitahara O, Furukawa Y, Tanaka T, Kihara C, Ono K, Yanagawa R Sinha SK, Chaudhary PM. (2004). Induction of apoptosis by X-linked et al. (2001). Alterations of gene expression during colorectal ectodermal dysplasia receptor via a caspase 8-dependent mechan- carcinogenesis revealed by cDNA microarrays after laser-capture ism. J Biol Chem 279: 41873–41881. microdissection of tumor tissues and normal epithelia. Cancer Res Sinha SK, Zachariah S, Quinones HI, Shindo M, Chaudhary PM. 61: 3544–3549. (2002). Role of TRAF3 and -6 in the activation of the NF-kappa B Lark AL, Livasy CA, Calvo B, Caskey L, Moore DT, Yang X et al. and JNK pathways by X-linked ectodermal dysplasia receptor. (2003). Overexpression of focal adhesion kinase in primary color- J Biol Chem 277: 44953–44961. ectal carcinomas and colorectal liver metastases: immunohisto- Smahi A, Courtois G, Rabia SH, Doffinger R, Bodemer C, chemistry and real-time PCR analyses. Clin Cancer Res 9: 215–222. Munnich A et al. (2002). The NF-kappaB signalling pathway in Lee SH, Shin MS, Kim HS, Lee HK, Park WS, Kim SY et al. (1999). human diseases: from incontinentia pigmenti to ectodermal Alterations of the DR5/TRAIL receptor 2 gene in non-small cell dysplasias and immune-deficiency syndromes. Hum Mol Genet 11: lung cancers. Cancer Res 59: 5683–5686. 2371–2375. Levine AJ. (1997). p53, the cellular gatekeeper for growth and division. Takakuwa T, Dong Z, Nakatsuka S, Kojya S, Harabuchi Y, Yang WI Cell 88: 323–331. et al. (2002). Frequent mutations of Fas gene in nasal NK/T cell Liang F, Liang J, Wang WQ, Sun JP, Udho E, Zhang ZY. (2007). lymphoma. Oncogene 21: 4702–4705. PRL3 promotes cell invasion and proliferation by down-regulation Tanaka H, Arakawa H, Yamaguchi T, Shiraishi K, Fukuda S, of Csk leading to Src activation. J Biol Chem 282: 5413–5419. Matsui K et al. (2000). A ribonucleotide reductase gene involved in

Oncogene XEDAR as a p53-induced regulator of anoikis C Tanikawa et al 3092 a p53-dependent cell-cycle checkpoint for DNA damage. Nature Weiner TM, Liu ET, Craven RJ, Cance WG. (1993). Expression 404: 42–49. of focal adhesion kinase gene and invasive cancer. Lancet 342: Tanikawa C, Matsuda K, Fukuda S, Nakamura Y, Arakawa H. 1024–1025. (2003). p53RDL1 regulates p53-dependent apoptosis. Nat Cell Biol Wu GS, Burns TF, McDonald III ER, Jiang W, Meng R, Krantz ID 5: 216–223. et al. (1997). KILLER/DR5 is a DNA damage-inducible Vogelstein B, Fearon ER, Kern SE, Hamilton SR, Preisinger AC, p53-regulated death receptor gene. Nat Genet 17: 141–143. Nakamura Y et al. (1989). Allelotype of colorectal carcinomas. Yan M, Wang LC, Hymowitz SG, Schilbach S, Lee J, Goddard A Science 244: 207–211. et al. (2000). Two-amino acid molecular switch in an epithelial Vogelstein B, Lane D, Levine AJ. (2000). Surﬁng the p53 network. morphogen that regulates binding to two distinct receptors. Science Nature 408: 307–310. 290: 523–527.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene